Re: [RE-SEND] [PATCH 3/4] s3c-fb: Add support EXYNOS4 FIMD

[Sylwester Nawrocki <s.nawrocki@samsung.com>]

On 06/15/2011 12:22 PM, Thomas Abraham wrote:
> On 15 June 2011 15:38, Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
>> On Wed, Jun 15, 2011 at 03:34:40PM +0530, Thomas Abraham wrote:
>>> Ok. Thanks for clarifying this point. In this case, maybe we could use
>>> the clk_add_alias API. With this API, we can have a instance of clock
>>> by name 'fimd' and then create a clkdev alias by name 'lcd' in the
>>> machine code. The driver can then continue to lookup with 'lcd' as the
>>> con_id.
>>
>> That's additional needless complexity.  If you don't embed the clk_lookup
>> structure in your struct clk but list out the dev/con/clk stuff separately
>> like OMAP does, you can easily assign the dev/con pairs to multiple
>> clks without hastle - and then it really doesn't matter what the clk
>> itself is called.
>>
> 
> Thanks for your suggestion Russell. The limitation in doing it the
> OMAP way is, in all Samsung machine code, all the clock instances are
> put into a array and then registered. So there is no separately named
> instance for each clock to build a dev/con/clk omap-type table.

I realize it might be quite a bit of work to convert the clock tables,
perhaps some scripts/trickery could be of help here.
I'm just wondering if this won't have to be done anyway when people
make agreement about common struct clk on ARM.

However there is a little problem that there are two clock arrays in
each Samsung platform, for clock disabled and enabled by default.
Not sure yet how to handle that cleanly.

Cheers,
Sylwester

-- 
Sylwester Nawrocki
Samsung Poland R&D Center